Too many factor to be considered
Due to the complexity of hydrodynamics around the membrane fouling phenomena, any finding in a specific condition may not be applicable to other conditions. For instance, higher steady state flux might be achieved with the fibers with smaller diameters in one condition, but it can turn out to be opposite in other condition depending on the length of the fiber used, air flow rate employed, MLSS of the mixed liquor, etc. Therefore, all the module design parameters should be considered holistically in the condition the module is aiming to be used. Following are the list of factors affecting membrane performance that need to be considered holistically to obtain optimum module configuration.
- Fiber dimension : ID, OD, and length
- Fiber property : permeability, pore size distribution, and flexibility
- Module structure : depth and width of fiber bundle, air nozzle position, etc.
- Aeration method : bubble size, air flow rate, bubbling frequency, etc.
- Mixed liquor property : Ranges of MLSS, SRT, viscosity, rheological property, etc.
The complex interactions among the parameters listed above can be found in literature. For instance, the fiber movement (or amplitude), which is a crucial factor affecting antifouling action of hollow fiber, significantly varies depending on the medium viscosity that is directly correlated with biological condition. As can be seen in Fig. 1, the correlation between fiber amplitude and fiber looseness is not consistent in all conditions even if fiber size, length and aeration rate are identical (Wicaksana, 2006). Depending on medium viscosity, the magnitude and the trend of fiber amplitude can change significantly. In a low viscosity medium (water, 1 cP), fiber amplitude continued to increase when fiber looseness increased from 0% to 5%. But, at high viscosity medium (water-glycerol mixture, 3.2 cP), fiber amplitude reached a plateau at a much lower amplitude at a looseness of 2%.
Fig. 1. Effect of fiber looseness and medium viscosity on fiber amplitude. Fiber length=70cm, OD=0.65 mm, ID=0.39mm, pore size = 0.2 micron, polypropylene, scouring air flow=2 L/min, nozzle size = 1mm (Wicaksana, 2006).
Contradictory phenomena are often observed from seemingly identical experiments depending on references due to the complex nature of hydrodynamics and membrane fouling phenomena that can hardly be controlled perfectly identically. The experimental results shown in Fig. 2 are all obtained with one 0.1 m2 flat sheet membranes manufactured by same company,Yuasa Co. (Japan), although the module used in a) was fabricated by Kubota and the other in b) was by Yuasa Co. The MLSS were close each other at 8 g/L and at 6 g/L for a) and b), respectively. The exact specifications of the diffusers were not known, but approximate diffuser specs are marked in the graphs (Fig.2). While the space between the membrane and the guiding wall is not known for a), it is 7 mm for b). Although experimental outlines are apparently very similar, observations were exactly opposite. One experiment (a) shows small bubbles are superior to large bubbles, but the other experiment (b) shows larger bubbles are superior. The causes are not clear, but it can be attributed to the subtle differences in mixed liquor properties, the unknown diffuser specifications and the flow channel dimensions, etc.
Contradictory observations also exist in hollow fiber membranes. Small bubbles (5 mm) were more effective than large bubbles (20 mm) in lab scale experiments performed with short membranes installed in a confined spaces of plat and frame module, where water was circulated by a pump and the bubbles supplied had no choice not to pass near the fibers (Fane, 2005b). However, lager bubbles generated from the diffusers with 6 mm and 8 mm pores are more effective than bubbles generated from diffusers with <3 mm pores in commercial hollow fiber modules, where actual bubble sizes are supposedly much larger than the diffuser pore sizes (Hong, 2010).
Membrane manufacturing method also affects membrane fiber properties even if basic polymer chemistries are identical. Thus the optimum module design can vary to some extent depending on manufacturers even if membranes are based on identical chemistry.
Most importantly, one module cannot be optimum for every situation at least in theory. Many of biology related factors are somewhat uncontrollable since they are depending on the wastewater flow rate, strength, composition, etc. Therefore, a module optimized for one condition is not necessarily the best for other condition.
a1) Coarse bubble
a2) Fine bubble
b) Various bubble sizes
Fig. 2 Effect of diffuser pore size on membrane fouling: a1) two coarse bubble diffusers with six 2 mm holes, a2) two fine bubble diffusers with unknown number of 0.5 mm holes (Sofia, 2004) b) diffusers with unknown number of various size holes (Ndinisa, 2006a). The experimental condition of a1) and a2): MLSS = 8 g/L, water temperature = 30 oC, gross flux = 16.9 LMH (or 0.48 m/d), and suction cycle = 8 min on/2 min off. The experiment in b) was performed at a MLSS of 6 g/L.
© Seong Hoon Yoon